Back to skill
Skillv1.0.0
ClawScan security
Lofy Projects · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignFeb 11, 2026, 9:35 AM
- Verdict
- benign
- Confidence
- medium
- Model
- gpt-5-mini
- Summary
- The skill's requirements and instructions are internally consistent for a local project-management helper, but a few runtime behaviors (meeting detection and proactive sending) are vague and could cause the agent to access external calendars or messaging channels if not constrained.
- Guidance
- This skill appears coherent for local project tracking: it wants to read and update a projects JSON file and produce prioritization and meeting-prep text. Before installing, confirm two things: (1) where meeting detection and 'send prep' actions will read/write (calendar, email, chat), and ensure you are comfortable granting access to only those specific services; (2) that the agent's file-write permissions are limited to the intended data/projects.json location (backup your files if needed). If you need the skill to send messages or access calendars, prefer explicit, auditable connectors (e.g., a calendar API token you supply with clear scopes) rather than letting the agent guess how to detect meetings. If you cannot get clarification from the author, treat proactive scheduling/sending as an operational risk and disable autonomous/background invocation for this skill.
Review Dimensions
- Purpose & Capability
- okName/description (project management, prioritization, meeting prep, time logging) match the instructions: read/update a projects JSON, compute priority scores, flag stale projects, and provide meeting prep. The skill does not request unrelated binaries, credentials, or config paths.
- Instruction Scope
- noteMost instructions are confined to reading/updating data/projects.json and producing recommendations. However, 'When a meeting is detected' and 'Send prep 2 hours before' are underspecified: they imply detecting external events (calendar/meeting system) and sending messages but do not declare how or where. That ambiguity could lead the agent to access calendars, email, or messaging integrations without explicit declared access.
- Install Mechanism
- okInstruction-only skill with no install spec and no code files — lowest install risk. Nothing is downloaded or written by an installer.
- Credentials
- okThe skill declares no required environment variables, credentials, or config paths. Its stated operations (local JSON read/write) are proportionate to the purpose. There is no unexplained request for secrets or external credentials.
- Persistence & Privilege
- notealways:false and default autonomous invocation are appropriate. That said, the instruction to 'proactively send prep 2 hours before' implies ongoing monitoring or scheduled triggers; achieving that behavior would require either external scheduler/cron or background invocation and possibly access to calendar/messaging — the skill does not request persistent presence explicitly.